AI agents have carried out more than 138 million cryptocurrency transactions through Coinbase’s x402 payment protocol, buying data, computing resources and access to AI model services, according to research from blockchain analytics firm Chainalysis reported by CoinDesk.
The activity offers an early look at a market in which software systems can spend money without a person approving each individual payment. The United States is currently the dominant identifiable source of payments for AI model inference, while China-linked wallets are more prominent in security and code-analysis services. However, the amounts involved are generally very small, and the transaction totals do not by themselves prove that all of the activity reflects genuine commercial demand.
U.S. wallets dominate model-inference payments
Chainalysis found that U.S.-linked wallets represented 88.1% of geographically identified payments for AI model inference. Chinese-linked wallets accounted for just 0.1% of those payments.
One possible reason, Eric Jardine, Chainalysis’ head of research, told CoinDesk, is that Chinese developers may be more likely to use open-weight models such as DeepSeek and Qwen. That could reduce their need to buy metered access to inference services, where users pay according to their consumption.
The research does not establish that open-source or open-weight AI is weakening the broader Asian market for blockchain-based payments. Jardine cautioned that the geographic difference should not be interpreted that broadly.
China-linked wallets showed a different concentration of activity. They represented about 90% of geographically identified payments for security and code-analysis services. Much of that spending was directed to AgentLisa, an AI-powered smart-contract auditing platform.
Stablecoins enable tiny automated payments
The x402 protocol, incubated by Coinbase and operating through its Base blockchain, allows AI agents to pay automatically for digital services using cryptocurrency. These services can include data, computing power and access to AI models.
More than 87% of the x402 payments analyzed by Chainalysis were worth between one-hundredth of a cent and 10 cents. Such amounts are generally too small to process economically through traditional credit-card networks, while stablecoins can transfer dollar-linked value in fractions of a cent.
This makes stablecoins a potential fit for machine-to-machine commerce. An AI agent can purchase a small piece of data or computing capacity as needed, without a human manually entering payment details. Stablecoins are crypto tokens designed to maintain a relatively stable value against an underlying reference asset, commonly a major currency.
The scale remains modest in financial terms. During one 30-day period in July, x402 processed roughly 75 million payments with a combined value of $24 million. The large number of transactions alongside the relatively low total value illustrates how much of the network’s activity consists of micropayments.
Chainalysis also identified signals that require caution when interpreting the growth. Among wallets first funded in the second quarter of 2026, 89.3% received additional funding within 30 days, compared with 70.6% for wallets first funded in the third quarter of 2025. The analysis excluded wallets that made payments on only one day.
The average interval between a wallet receiving funds and making its first x402 payment also fell from 73 days in late 2025 to 18 minutes in the third quarter of 2026. These figures show faster participation and repeat funding, but transaction counts alone cannot determine how much activity represents sustainable business demand.
Payments industry watches agentic commerce
The findings arrive as Visa, Mastercard, Stripe and other payments companies have backed the x402 protocol. Coinbase has also started allowing business customers to accept payments from AI agents.
The broader idea is that AI systems could become independent economic participants, with their own wallets and the ability to select and pay for digital services. Chainalysis described each agent as effectively a new blockchain user in its report.
Yat Siu, co-founder and chairman of Animoca Brands, has estimated that as many as 100 billion AI agents could be transacting online over the next several years and using cryptocurrency. Cloudflare Chief Strategy Officer Stephanie Cohen has separately said stablecoins are especially suited to the types of transactions AI agents are expected to make initially. Those statements are projections and views, rather than measurements of current activity.
The Chainalysis figures cover identifiable transactions through x402, not the entire global market for AI-related payments. For now, the data points to an emerging system in which autonomous software is beginning to make frequent, low-value purchases, with U.S.-linked wallets leading spending on paid AI model inference.
Moneyiar market picture
Moneyiar market data at 21:04 on October 15, 2026, showed Chainlink (LINK) at $12.4749950764, up 0.94% over 24 hours and down 1.41% over seven days. JUST (JST) was $0.1323151494, down 6.32% in 24 hours and 4.12% over seven days.
OriginTrail (TRAC) stood at $0.4254657292, down 2.88% over 24 hours but up 6.21% over seven days. ORDI was $3.9537316801, down 0.51% over 24 hours and 5.58% over seven days. These figures are Moneyiar data at the stated time and are separate from Chainalysis’ x402 transaction research.
